Orchestrate a swarm of coding agents across Claude Code and Codex
The author built a containerized Claude Code plus Codex setup with an MCP letting Claude spawn Codex sub-agents and sub-orchestrators, plus a self-hosted git/CI stack, in 24 hours. ·
Done withClaude CodeCodexHermes Agent
- What they did
- The author used a Claude model they call Fable to build a containerized Claude Code plus Codex package with a custom MCP. Through it Claude calls Codex sub-agents, including multi-level sub-orchestrators, so the Codex subscription gets used too. They also had it build a self-hosted CI/CD pipeline deploying to AWS, replacing GitHub Actions after free credits ran out. Bugs are logged as git issues and fixed by the agent.
- How it went
- Over 24 hours the author says they got the full stack working, tested it themselves and with their wife, and are close to shipping several SaaS products. No metrics are given, and the claims are self-reported.
- Worth knowing
- The author says the model is expensive and wishes it were cheaper; free GitHub Actions credits ran out quickly, which prompted the custom CI/CD.
Try it yourself with Claude Code
Help me set up a multi-agent coding workflow for [project description] where you act as the orchestrator and delegate subtasks to Codex as sub-agents, each working in its own container or git branch. Give me a short plan first, then build a minimal version that completes one test task end to end and reports what each sub-agent did. Pause before pushing to any remote or deleting any branch.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Nov 12, 2026
Related use cases
Turn tickets into merged pull requests
Pulls tasks from GitHub Issues or Linear, then runs Claude Code or Codex in isolated Kubernetes pods to carry each one through to a PR.
Guide a music video project from concept to deliverables
A Hermes agent walks a creator through an 8-step workflow producing lyrics, Suno prompts, image and video prompts, and YouTube metadata.
Scan an app for UX improvements with computer use
A user tried Claude Code's computer-use feature on Windows to scan their app for UX improvements, and asks whether it works well there.
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Review GitHub pull requests automatically
An n8n AI agent triggers on each new PR, fetches diffs, calls an LLM for a review, and posts comments and labels back on the PR.
Migrate a website from a CMS to raw code and Markdown
Lee Robinson used Cursor agents to move cursor.com off its CMS in three days for $260 in tokens, with subagents handling a broad refactor.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.